Re: [BUG] -T html: .RS/.RE incorrectly indented

[G. Branden Robinson]

Hi, Alex!

At 2022-01-24T21:48:51+0100, Alejandro Colomar wrote:
> I wrote a filter chess game a few days ago, and wrote some manual
> pages for it.  When I decided to serve the manual pages online on my
> website, I found a few bugs, that I guess are groff responsibility,
> since I produced the html pages this way:
> 
> $ <chess-init.6 groff -m man -T html >chess-init.6.html

Yup, nothing wrong with this.  (Though as you probably know, most people
don't perform input redirections before command words in the shell.  It
does have the piquant effect of making the input _look_ superficially
more like HTML itself...)

> I'll report one of them here, and expect a few more email threads for
> other bugs.

Full speed ahead!  The list's been quiet this month.  (bug-groff@, not
so much.)

> So, the first problem I met is, for the following code:
> 
> [
> .PP

This `PP` call is superfluous in light of the `TP` immediately after it.
However, it doesn't seem to have any effect on the bug.

> .TP
> Board:
> .RS
> .TP
> .B \(aq \(aq
> Empty black cell.
[snip]

> But, on the html file, I see (firefox and chrome show the same thing):
> 
> [
>        Board:
> 
>        ' '           Empty black cell.
> 
>        '-'           Empty white cell.
> 
>        Pieces:
> 
>               B, b   Bishop.
> 
>               K, k   King.
> 
>               N, n   Knight.
> 
>               P, p   Pawn.
> 
>               Q, q   Queen.
> 
>               R, r   Rook.
> ]

Can reproduce with Firefox.

> With the first block incorrectly indented.  Text-based browsers such
> as w3m(1) or links(1) also show weird indentation; you can check them
> too.

With lynx(1) I get no indentation at all from the relative insets, so
it's inconsistent, but also invisible.

> I haven't seen any browser that displays the proper indentation,
> so I guess it's not a browser problem (but it might very well be).

I'm not sure, but I lean against it.  grohtml(1), which I don't fully
understand, uses an HTML table to represent essentially the entire body
of this document.

It does that even if I comment out the RS/RE calls.  I guess grohtml
imposes a table on the document body unconditionally.

> You can check the manual page full source code at
> <http://www.alejandro-colomar.es/src/alx/alx/games/chessutils.git/tree/share/man/man6/chess-init.6>
> and the html page at
> <http://www.alejandro-colomar.es/man/man6/chess-init.6.html>

I see no problems with your input apart from the aforementioned harmless
redundancy, but then I expect such high quality from someone who's
ground through as many pages as you have.  :D

I think you've identified a legitimate grohtml bug.  Could you file it
with the Savannah bug tracker
<https://savannah.gnu.org/bugs/?group=groff&func=additem>?  I'd title it
"grohtml does not track man(7) relative insets accurately" or something
like that.

Regards,
Branden

signature.asc
Description: PGP signature